DNA testing for sickle cell anemia in Africa: Implementation choices for the Democratic Republic of Congo

Abstract Background Hemoglobin‐based tests form the reference diagnostic test for SCA. In limited resource countries, these tests face limitations including cost, low sensitivity due to recurrent transfusions in endemic malaria region, and interference from fetal hemoglobin in neonatal diagnostic. This study aimed at adapting DNA‐based SCA tests to limited resource countries and evaluating the economic benefit. Methods 338 participants were recruited in the Democratic Republic of Congo, sorted in 3 cohorts based on venous blood, umbilical cord blood (UCB) and buccal swab sampling. RFLP was performed to identify mutated allele. The feasibility and technical validity of this RFLP was evaluated for specimens collected on DBS cards and on EDTA tubes. RFLP on DBS stored at room temperature was regularly repeated to assess sample conservation. Finally, the cost analysis was performed. Results DBS cards yielded identical results to extracted DNA. Repeated testing returned the same result after four years. The DBS‐based test performed on UCB or on buccal swab had a sensitivity and a precision of 100%. Cost comparison indicated that our approach costs half price of the widely used isoelectrofocussing of hemoglobin. Conclusion The implemented DNA‐based test approach overcomes the limitations faced by hemoglobin‐based tests, while being more affordable. We propose to implement the RFLP test as a first line diagnostic test after transfusion and as second tiers for newborn screening. However, users should be aware that this test is unable to differentiate HbC from HbS or identify other point mutation of gene deletion of HBB gene.


| INTRODUC TI ON
Low-income countries such as the Democratic Republic of Congo (DRC) suffer a heavy burden of sickle cell anemia (SCA) with about 2% of newborns being homozygous for the classical SCA mutation, and 17%-24% being heterozygous carriers. 1,2 The disease-related mortality remains high in many sub-Saharan African countries as 50% to 90% of affected children die before 5 years of age, 3 often before being formally diagnosed. Thus far, nationwide newborn screening is not available in DRC.
The most commonly available SCA tests in many localities in DRC include the Itano solubility test and the Emmel sickling test.
These tests detect only the presence of hemoglobin from the abnormal allele (HbS) and are unable to detect the hemoglobin from the reference allele (HbA). Therefore, these tests are unable to distinguish sickle cell trait (AS profile from heterozygotes individuals) from SCA (SS profile from homozygotes patients). 4,5 Thus, they are of limited diagnostic and clinical value. Electrophoresis on acetate cellulose and isoelectrofocussing (IEF) has been introduced in DRC since 1954 and 2009, respectively. 1,6 However, these technologies are not widely available across the country. Samples need to be shipped from rural areas to the laboratories located in urban areas.
Such sample transfer and storage represent big challenges and may impact on the cost and quality of the test. These technical and logistic challenges as well as the cost of the test should be considered when implementing a diagnostic test for SCA in a resource-limited setting.
In addition, the hemoglobin electrophoresis-based techniques have low accuracy during the neonatal period because of the presence of fetal hemoglobin, or after a transfusion because of the presence of donor's hemoglobin. 7 Studies on Congolese SCA patients have reported that anemia is the most common sign and often the first manifestation. 8,9 Thus, when SCA is suspected, patients have usually a recent history of transfusion, forcing practitioners to delay the diagnostic test for 3 months. Unfortunately, because of other intercurrent factors, such as malaria and other infections, that are very frequent in this setting, it is common for these children to return to the hospital and receive another transfusion within these 3 months' window, further delaying SCA confirmation by hemoglobin electrophoresis-based methods. Therefore, the optimal diagnostic test for SCA should retain high accuracy regardless of recent transfusion or in neonatal period.
Previous studies have shown that DNA-based tests are not affected by transfusion or fetal hemoglobin. 10

Samples collection and transportation
For each participant, 4 ml of peripheral venous blood was collected in a EDTA-coated tube (BD Diagnostics). Before removing the sampling needle, a drop of blood (approximately 50 µl of blood) was deposited on a circle of a FTA (Flinders Technology Associates) Elute card (Whatman WB120206). This type of FTA card is chemically treated filter papers that lyse cells, denature proteins and stabilize nucleic acids on the cellulose fibers ( Figure 1A) The spotted FTA cards were kept at room temperature for 90 minutes allowing the sample to dry prior to transportation in an individual plastic bag, adequately labeled. In order to reduce the cost of the experiment, each individual FTA card was divided into two pieces, allowing one card to be used for two patients ( Figure 1A).
The spotted FTA cards (dried blood spot, DBS) were stored in zip bags at room temperature on a clean shelve until being used for applications as further described below. coshg.org).

DNA extraction and molecular test
The venous blood collected in EDTA tubes was used for DNA extraction by salt-induced precipitation of cellular protein method also known as standard salting out procedure. 11 The extracted DNA was then re-suspended in 300 µl of TE buffer and kept at −20°C until the analysis. The standard manual salting out extraction procedure lasts around 24 hours. In our study, DNA extraction was performed within two days following the sample collection.
A restriction fragment length polymorphism (RFLP) test for SCA was implemented in our laboratory as previously described. 12 This test uses the DdeI restriction enzyme to differentiate between the mutant allele (S) and the wild-type allele (A) on a PCR product spanning the SCA mutation site ( Figure 1B).

| Cohort 2. Umbilical cord blood Study population
Umbilical blood cord was collected from 102 newborns in two ma-

Samples collection and transportation
Umbilical cord blood (UCB) was obtained "in utero," meaning before the placenta was expelled, the fetal extremity of the umbilical cord was disinfected, clamped, and sectioned. Blood sample was collected by gravity into a 4 ml EDTA tube. A drop of UCB was directly spotted onto two circles of a FTA card. Samples were transported to the laboratory of the Center for Human Genetics of the Kinshasa University at room temperature packed in plastic zip bags for spotted FTA cards and in isotherm boxes for UCB in EDTA tubes.
The UCB sample collected in EDTA tubes was used for DNA extraction following the standard salting out procedure, whereas spotted FTA cards were used as described below.

| Cohort 3. Saliva Study population
A group of 70 participants was recruited, among which 20 were from the CMMASS and 50 from the University Hospitals of the University of Kinshasa (CUK). Previous molecular diagnosis using DNA extracted from peripheral blood returned homozygous mutation genotype (SS hemoglobin type) in 43, heterozygous (AS hemoglobin type) in 13, and homozygous wild type (AA hemoglobin type) in 14.

Samples collection and transportation
For each participant, buccal cells were collected using sterilized cotton buds made with wooden sticks. Patients were recommended not to eat, drink, smoke, chew gum, neither use toothpaste nor mouthwash during the 30 minutes preceding sample collection. Samples were collected by scratching while rotating the swab on the mucosa of cheeks for 30-60 seconds. After sampling, the swab was scratched against a circle of FTA elute card for 10 seconds in order to transfer the cells from the swab onto the FTA card.

DNA testing
Each sample collected in EDTA tubes underwent DNA extraction by manual salting-out method (SO). 11 Extracted DNA was later used for a PCR reaction and the RFLP using an Applied Biosystems 2720 Thermal Cycler. We refer to this protocol as SO-RFLP across the article.
For samples collected on FTA cards, a DNA extraction step was not performed. One 1.2 mm disk was punched from the spotted FTA card, washed with purified water in a 2 ml reaction tube as recommended by the manufacturer, then transferred into the PCR reaction tube and mixed with the PCR reaction mix ( Table 1). The PCR product was used for enzymatic restriction as described below. In order For each run, two controls were used, including a No DNA control (disk punched from blank unspotted paper) and a positive control (disk punched from FTA card spotted with blood sample of a Congolese SCA patient with known genetic profile). The composition of the PCR mix is presented in Table 1.
Either 1 µl of DNA extracted by SO (100-300 ng) or a 1.2 mm washed FTA punch was added into the PCR mix for the SO-RFLP or FTA-RFLP, respectively.
The PCR program consisted of an initial denaturation step at 95°C for 5 min, followed by 32 amplification cycles, each cycle included a quick denaturation at 95°C for 30 s, annealing at 58°C for 30 s, and extension at 72°C for 30 s. The reaction was concluded with a final extension step at 72°C for 5 min.
The PCR product was digested using the DdeI enzyme (Promega Corporation, Belgium, Cat: R6291) according to a standard protocol.
The DdeI enzyme has two restriction sites (ctgag/) within the PCR product ( Figure 1B). The digestion products were visualized on 2% agarose gel ( Figure 1C). A non-mutated or wild-type allele produced three fragments with 201 bp, 167 bp, and 72 bp size, respectively.
Conversely, on an allele with the S mutation (mutated allele), the first restriction site is abolished, resulting in only two fragments of 368 bp and 72 bp, respectively. For homozygotes SS individuals, only these two bands appear, three bands are observed for homozygotes AA, whereas four bands are identified in heterozygotes AS, that is, 368 bp, 167 bp, 201 bp, and 72 bp bands.

Sensitivity and precision
Sensitivity, considered as the ability of the test to identify variants that are present in a sample, was calculated using the formula TP/TP + FN. The precision, defined as the fraction of variant calls that match the expected, reflecting the number of FP per test, was determined using the formula TP/TP +FP. 13 As previously demonstrated, the sensitivity threshold is influenced by the prevalence of the disease in the studied population. 14 For the previously reported incidence of 1.4% in DRC, 1 an error rate of 5% will result in multiple children being denied timely care. Therefore, we set the minimum threshold at 99%.

Evaluation of repeatability and reproducibility
Repeatability, defined as the percent agreement between the results of successive tests carried out under the same conditions of measurement, 13 was assessed in two ways: the intra-run and inter-run repeatability. Five FTA cards spotted with venous blood were randomly selected. Three disks from each in five cards were included a run to evaluate intra-run repeatability. This was repeated three times to assess inter-run repeatability.
We also assessed the reproducibility, which evaluates whether changes in testing platforms, reagent supplies, and operators cause significant change in test results. A second operator tested one disk from the test set of five FTA cards using a different PCR thermocycler. Reproducibility was calculated as the percent agreement between the results of tests under different conditions. 13 Repeatability and reproducibility were assessed for UCB and saliva spotted on FTA cards using six randomly selected cards for each specimen and running the same experiment as described for the venous blood. The considered cut off was 90% (Clinical Laboratory Improvement Amendments CLIA).

Stability of spotted FTA cards in local conditions
Each FTA card was stored in an individual plastic zip bag at room temperature, in a clean closet, whereas the extracted DNA was stored at −20°C. We randomly selected 15 FTA cards from study group 1.
The selected FTA cards were repeatedly tested every 6 months by FTA-RFLP as described above. Thus far, we have completed testing after 4 years of conservation.

| Data analysis
The SO-PRC-RFLP was considered as the gold standard to which results from the FTA cards were compared. The FTA-RFLP method was compared to the gold standard regarding the cost and conditions. Sensitivity and precision were computed. Statistical analysis was performed with SPSS version 21. Kappa statistics was used to evaluate the agreement between two different testing methods.
We also assessed the stability of FTA-spotted venous blood samples after 4 years of storage at room temperature.

| Ethical compliance
This study was compliant to international ethics laws and regulations. Prior to the inclusion, parents and patients were fully informed and provided signed informed consent/assents. The study protocol  there was a perfect agreement between FTA-RFLP and SO-RFLP (kappa = 1; Figure 2).

| FTA-RFLP on umbilical cord blood
The 102 FTA cards containing UCB samples were tested by FTA-RFLP and returned 1 SS, 14 AS, and 87 AA. Later, the genomic DNA extracted from the 102 UCB samples collected in the EDTA tubes was also tested by SO-RFLP. This returned the exact same result for each participant, corresponding to a perfect agreement between the two techniques (kappa = 1; Figure 2).

| FTA-RFLP on buccal swab samples
The

| Sensitivity and precision
The results from venous blood collected on FTA cards, UCB spotted on FTA cards and swabs rubbed on FTA cards, all returned identical results to the respective outcome obtained with the SO-RFLP. No F I G U R E 2 Study algorithm and results. The top row presents the five institutions where participants were recruited; the second row shows the composition and specimens for the three cohorts; the third row indicates the two testing approaches applied to each of the cohort. The distribution of the genotypes obtained by either of the testing approaches for each cohort is presented at the bottom of the figure false positive or false negative result was observed. Therefore, the sensitivity as well as the precision was 1 for DNA-based SCA test using either of those three specimens.

| Repeatability and reproducibility
Intra-and inter-run replicates returned the exact same profiles for each of the randomly selected individuals. Moreover, the change in the operator and the PCR thermocycler did not alter the outcome of the test.

| Stability of spotted FTA cards in local conditions
The results of 4 years of storage indicated that the profile remained clearly recognizable, suggesting the DNA quality was not altered by the local storage conditions: shelve storage in a dust-free closet at room temperature (between 25 and 35 Celsius degrees). The results remained identical over the years up to 4 years when the last test was performed (Figure 3).

| Conditions and cost comparison
The conditions and the cost of both tests are presented in Table 2.
The cost of the SO method was based on estimation by Chacon-Cortes et al. 15 While the personnel cost for both SO-RFLP and FTA-RFLP was estimated according to the average monthly salary of a laboratory technician in DRC. Due to the ease in sample collection, transport and handling, the use of DBS cards such as FTA cards is gaining in acceptance for genetic analysis in multiple fields including infectious diseases, agriculture, and pharmacogenetic. [16][17][18][19][20][21] The RFLP was previously identified as a reliable and cheap approach for DNA-based diagnostic of SCA. 7 However, users should note that this test cannot differentiate HbS from HbC since both mutations affect the same restriction site and result in the same profile on agarose gel electrophoresis. Also, this test is not designed for the identification of others mutations that may cause SCA phenotype, such as compound heterozygous HbS/βthal or HbS/HbO.

| DISCUSS ION
Fortunately, such mutations are rare in the DRC and absent from this cohort (based on sequencing data, not included in this article).
The FTA cards were tested as an alternative to the EDTA tube sampling and DNA extraction by salting out. In this study, we obtained the same genomic profile using a 1.2 mm FTA disks or 1µl of DNA extracted by salting out from the same participant. We con- labor-intensive and time-consuming. This resulted in a high personnel cost that made a significant difference between the two approaches. Interestingly, the estimated cost for SO-RFLP is also half of the price commonly charged in Kinshasa for the IEF. Hence, our proposed approach of FTA-RFLP is not only more accurate than IEF, but it is also more efficient on challenging specimens such as UCB, it is suited for non-invasive testing using a buccal swab, and, most importantly, four times cheaper than the IEF. Therefore, switching to DBS DNA-based test as the confirmatory approach in the newborn screening program, and as first-line test for SCA diagnostic in other aged groups will significantly reduce the financial burden while offering the highest accuracy. This approach can also be offered in mass campaigns as a non-invasive method.
In addition, this test can be used as first-line test for screening sickle cell trait in blood donors and blood donations. This will increase blood transfusion safety. 28

| CON CLUS ION
The study aimed at implementing a DNA-based SCA test in resourcelimited settings and evaluating the economic benefit. We have shown that this procedure is feasible in resource-limited settings, retains high accuracy for various sample types, including venous blood, umbilical cord blood, and buccal swab, while remaining affordable. We recommend this test as a first-line diagnostic test after transfusion as well as in blood donors and blood donations, and as second line for newborn screening.

ACK N OWLED G M ENTS
The authors are grateful to patients, parents, and the families for agreeing to participate in this study. We thank Bruno Vankeirsbilck of UZ Leuven Molecular Laboratory and his team for the technical and logistic assistance.

DATA AVA I L A B I L I T Y S TAT E M E N T
We included all the data related to this research in the article.
Moreover, all data are available from the corresponding author.